The human body hosts a tremendously diverse microbial community on external and internal host surfaces such as the skin, respiratory tract, mouth, gastrointestinal tract, reproductive tract, urinary tract and eyes. The total number of microbial cells (1014) exceeds the number of host cells (1013) with an order of magnitude and comprises many thousands of species. Their enormous enzymatic diversity, their capacity to trigger host immunological responses and the possibility of modulating physiological processes within the host that are involved in the etiology of cancer, obesity, or cardiovascular diseases, makes these host-associated microorganisms a confounding factor in the general health-status of an individual.
More in particular, the gastrointestinal tract, which is still an environment external to the host's body, is a site where the microbial involvement in host processes is highly significant. Up to now, many studies have targeted the microbial community in the gut lumen by investigating their fermentation activity, metabolic potencies and changes in community composition due to certain treatments. However, specific microbe-host interactions often depend on the ability of microorganisms to adhere to the gut surface. Previous studies employed cell culture experiments such as Caco-2, T84, HT29 etc. to investigate the specific adhesion of probiotic or pathogenic microorganisms to epithelial cells. Yet, for reasons of cytotoxicity, cell cultures are very sensitive to co-incubation with mixed microbial slurries, thus limiting the experimental time to 2 hours maximally. Moreover, the growth of cell cultures is time-consuming and it is not possible to perform high-throughput screenings to evaluate many bacteria or components at the same time.
It has been proposed that the structure and composition of the gastrointestinal ecosystem reflects a natural selection at both microbial and host levels. Such ecosystem requires a series of evolved, nested equilibria to achieve the overall homeostasis, in analogy with the Evolutionary Stable Strategy theory. Host-derived signals and microbial-derived signals may be either unlinked or linked. Linked signals imply a selective pressure based on and favoring co-evolution (Blaser and Kirschner, 2007, Nature, 449: 843-849). The above mentioned natural selection primarily occurs along mucosal surfaces by a host-microbiota cross-talk that leads to the modulation of the host immunity.
Specific immunological signalling pathways may be triggered upon recognition by the host of microbial cell wall patterns. Microorganisms have a wide variety of MAMPs (Microbial Associated Membrane Patterns), which can be recognized by specific host receptors on epithelial cells. An example of this recognition process is the family of Toll Like Receptors (TLR), which recognize specific microbial adhesins such as lipopolysaccharides (LPS), lipoteichoic acids, fimbriae, pili etc. However, in the intestinal epithelium, where the interaction between the host and the endogenous microbiota is the highest, TLR can be downregulated. Prior to microbial invasion of epithelial cells, microorganisms need to reach the epithelium. However, there is no direct contact between luminal bacteria and the epithelial surfaces. The actual surface that bacteria encounter when approaching the host epithelium, is the mucus layer covering the epithelium. The study of the microbial ability to adhere to the intestinal mucus layer is therefore an important prerequisite when assessing a microorganism's ability to interact with the host. Although it is possible to trigger mucus production in for example HT29 (Novellvaux et al., 2006) as well as MKN1, MKN7 and MNK45 cell-lines (Linden et al., 2007), it is not possible to monitor colonization and persistence of mixed microbial communities on these cells for reasons of cytotoxicity. In vitro models allowing the study of microbial mucus colonization over a longer time-frame are therefore much needed.
The main constituents of mucus are mucins, which are usually present at a concentration ranging between 2% to 10% (w/v). These are glycoproteins containing high proportions of carbohydrates—usually between 70% to 85% (w/w). The mucins are unusual glycoproteins in that most of the carbohydrate side-chains are linked to the protein at serine and threonine residues via an oxygen atom (i.e., they are “O-glycosylated”), although N-glycosylation also occurs. The structure of a typical mucin molecule consists of a protein to which carbohydrate side-chains are linked by O-glycosylation and/or N-glycosylation. The protein backbone consists of several thousand amino-acid residues (in the case of MUC2, a major type of mucin found in the gastro-intestinal system (GIT), the number of amino acids is 5,179) and contains regions with many oligosaccharide side-chains and other regions without such side-chains. The oligosaccharide-rich regions are resistant to proteases, whereas the other regions are protease-sensitive. The oligosaccharide-containing regions of the protein are rich in serine, threonine and proline. The side-chains usually consist of two to twelve residues from a restricted range of sugars—usually galactose, fucose, N-acetylglucosamine, N-acetylgalactosamine, mannose and sialic acids.
The protein backbones of the various mucins produced are encoded by a large family of MUC genes. Which of these genes are expressed depends on the particular body site. Hence, MUC2 is the main type of mucin protein produced in the intestinal tract whereas the expression of six MUC genes has been detected in the cervix with different patterns of expression at different phases of the menstrual cycle. In the oral cavity, two well-known secreted mucins, mucus glycoprotein 1 (MG1) and mucus glycoprotein 2 (MG2), have been identified. High-molecular-weight MG1 consists mainly of the MUC5B gene product (Nielsen et al., 1997), whereas the low-molecular-weight MG2 is a product of the MUC7 gene (Bobek et al., 1993). However, only a few investigators have reported expression of membrane-associated mucins in the human oral cavity. Sengupta et al. (2001) reported that MUC1 was expressed in the ducts of the minor salivary. Liu et al. (2002) showed that MUC1 and MUC4 but not MUC3 and MUC13 were expressed in the human parotid and submandibular glands. Recently, it has been shown that cultivated stratified human oral mucosal epithelial sheets express the transcripts for the membrane-associated mucins, MUC1, MUC4, and MUC16 but not MUC3, MUC12, MUC13, MUC15, and MUC17 (Hori et al., 2007; Hon et al., 2008).
Differences also exist with regard to the glycosylation patterns of a particular type of mucin protein—hence the composition of the carbohydrate side-chains of a mucin with a MUC 1 protein backbone will be different in the respiratory and intestinal tracts.
Besides mucus adhesion, some bacteria can invade the mucus layer and utilize mucin as Carbon, Nitrogen and energy source. Mucin polymers need to be hydrolyzed, prior to the assimilation of mucin oligomers, mucin monomers and amino acids. The structural complexity of mucin polymers entails that the complete degradation of them by a single microbial species is unlikely. Such degradation requires the production of a whole range of enzymes in a certain order because regions of the molecule only become accessible once others have been removed. This is more readily accomplished by microbial consortia rather than by individual species. The range of enzymes needed to achieve complete degradation of mucin is shown in Table 1.
TABLE 1Enzymes required for complete mucin degradationType of enzymeRole in mucin degradationSulphatasesRemoval of terminal sulphate residues, therebyexposing underlying sugars rendering them moresusceptible to the action of glycosidases.SialidaseRemoval of terminal sialic acid residues: this exposes(neuraminidases)underlying sugars to the action of glycosidases: thesialic acid itself can be further degraded byacetylneuraminate pyruvate lyase to N-acetylmannosamine which can be used as Carbon andenergy source by some bacteria.ExoglycosidasesCleave sugars from side chains (e.g. β-Dgalactosidase, N-acetyl-β-D galactosaminidase,a-fucosidase, N-acetyl-β-D glucosaminidase).EndoglycosidasesCleave entire side-chain from the peptide backbone orattack the side-chain at sites other than the terminalresidue - this may occur before or after the side-chainhas been cleaved from the protein.Peptidases/Cleave at non-glycosylated regions: degrade proteinproteasesbackbone after side-chains have been removed.
The ability to degrade mucin, entirely or partially, has been detected in microbes or microbial consortia inhabiting all mucosal sites of the body. The removal of carbohydrates and other components from the glycoprotein compromises the protective function in the gut, especially when the rate of mucus breakdown exceeds the rate of mucus production. The composition and metabolic activity of mucosal microbial communities is quite different from the luminal microbial communities in the gut. Given their ecological significance regarding gut microbiota composition and their putative role in inflammatory bowel disease such as ulcerative colitis, it is crucial to understand the colonization, composition and metabolic activity of the gut mucosal microbial population.
Once gut microorganisms can adhere to the mucosa, they have the chance of forming a biofilm on the mucosa within the timeframe that the host epithelium renews its mucus layer. Biofilms are considered to reflect the most common steady-state for bacterial growth (Costerton, 1995) with the gut microbial biofilm constituting a driving factor in the establishment and maintenance of the spatially diversified microbial community (Hooper and Gordon, 2001). A special feature of mucosal biofilms is the presence of oxygen at the base, due to diffusion from the host blood stream across the epithelium. Luminal oxygen concentrations in the colon can even rise to 30 mm Hg. This would compromise the colonization and growth of strict anaerobes such as sessile Fusobacteria which are found to have an important “bridging” function within biofilms, forming co-aggregation/co-adhesion bridges between early colonizers and late colonizers and thus contributing towards biofilm establishment and accumulation (Kolenbrander et al., 2000). Yet, the growth of Fusobacteria in the mucus biofilm is tenable because of their local association with aerobes and facultative anaerobes, which locally deplete the mucus layer of oxygen. The presence of oxygen in the mucus layer also allows the production of reactive oxygen species such as for example O2—, H2O2 and OH from bacterial origin. For example, Enterococcus faecalis has been shown to form hydroxyl radicals by aromatic hydroxylation, thus being able to elicit oxidative stress towards the epithelium (Huycke and Moore, 2002).
Different in vitro approaches have been used to evaluate the influence of substances and organisms on the gastrointestinal system and its flora including both batch and continuous culture systems.
Batch Type Fermentors
Short term batch culture systems allow a rapid screening and a flexible design to assess for instance the interindividual variability. However, in this simplified environment, the control of changing conditions is not possible and only short term experiments can be conducted.
Single-Stage Continuous Fermentors
Single-stage continuous fermentors offer a better model for specific regions of the GI tract under controlled conditions. Nevertheless, stability of the microbial community under long term studies is not always possible.
The Reading Simulator
The Reading simulator (Macfarlane and Macfarlane, 2007) simulates the gut using a 3 stage continuous culture with three glass vessels (220 ml, 320 ml and 320 ml) and different pH in each vessel (5.8, 6.2, and 6.8), mimicking the human proximal, transverse, and distal colon, respectively. Each vessel is inoculated with 100 ml of 20% (w/v) of human feces. The system is run for 14 days in order to achieve a steady-state condition in the vessels, then for 3 weeks to test a specific compound and finally, for a washout period (2 weeks) to determine how long the changes induced by the test substance can still be measured in the absence of the substrate itself.
The EnteroMix Colon Simulator
The EnteroMix model has four parallel units each comprising four glass vessels, allowing four simulations to be run simultaneously using the same fecal inoculum (Makivuokko et al., 2006). EnteroMix model vessels 1, 2, 3, and 4 have small working volumes (6, 8, 10, and 12 ml, respectively). The pH levels are controlled. The simulation begins with three hours of incubation of 10 ml of the fecal inoculum and then 3 ml of fresh simulator medium with (three test channels) or without (one control channel) test substance is pumped to the first vessel. The medium is fermented in the first vessel for three hours, after which 3 ml of the fermented medium are transferred to the second vessel, and 3 ml of fresh medium are pumped to the first vessel. The fermentation lasts for 48 hours, after which samples are collected from each vessel and the simulation is terminated.
In the two latter systems the upper part of the GI tract (stomach and small intestine) is completely absent. Moreover, in the EnteroMix colon simulator, the volumes are small when compared with the in vivo situation, there is no stabilization of the microbial community and only short experiments can be performed.
In Vitro Gastrointestinal System Models (GIT)
Currently there are two complete in vitro GIT models available in which studies can be performed and labor, time and costs are reduced when compared with in vivo studies, without ethical constraints. In these two systems, the kinetics of the gut are simulated by controlling the concentrations of gastric, small intestinal and pancreatic enzymes, bile, pH, temperature, feed composition, transit time in the GIT and the anaerobic environment with physiological relevance.
TIM 1 and 2 (TNO Intestinal Model)
The first in vitro GIT model is TNO's gastrointestinal model (TIM). This model actually comprises two complementary parts; the systems 1 and 2 introduced by Minekus et al. in 1995 and 1999 (US55255305 and Minekus et al., 1999) and WO1994009895. The TIM 1 system contains four computer-controlled chambers simulating the conditions of stomach, duodenum, jejunum and ileum. The TIM 2 system consists of four glass modules mimicking the proximal colon of monogastric animals. In these dynamic models fluid transportation from vessel to vessel happens via peristaltic valve-pumps and there is constant, although passive, absorption of water and fermentation products through dialysis membranes. For the simulation of intestinal absorption TIM 1 has two integrated 5 kDa dialysis membranes, next to jejunal and ileal modules. TIM 2 has one hollow-fiber membrane (molecular mass cut-off value 50 kDa) in the luminal part en of the system. The pH-values are monitored in each compartment. In a TIM 2 simulation the model is inoculated with 200 ml of fecal inoculum. Microbiota are allowed to adapt to the conditions for 16 hours, however there is no long-term stabilization of the microbial community and the volumes in the different chambers are small when compared with in vivo situations.
The Simulator of the Human Intestinal Microbial Ecosystem (SHIME)
The second in vitro GIT model is the SHIME. The conventional SHIME is a dynamic model of the human gut comprising 5 compartments respectively simulating the stomach, small intestine and ascending, transverse and descending colon. The stomach and small intestine compartments mimic the enzymatic and physicochemical environment by controlling pH and residence time and the dosing of a proper nutritional medium, enzymes and bile salts. By controlling the pH, redox potential and residence times, the different colon compartments each harbor a microbial community that corresponds to that of the in vivo situation in terms of metabolic activity and community composition. In this model a typical stabilization period of three weeks and a basal period of two weeks are followed by treatment and wash-out periods.
Biofilm Simulation Model (Lacroix, Switzerland)
All the models presented until now do not take into account an important aspect in the GI tract: adhesion of microorganisms to the mucus layer, biofilm formation and its potential role on the host physiology and structuring of the microbial community and on cross-talk. Patent applications US2005186633 and US20040101906 from Lacroix et al. (2005) addressed the issue of the biofilm formation in the GI tract. They claimed a system utilizing cell immobilization in anaerobic continuous-flow cultures for modeling GI system. Microbes from fresh fecal samples are immobilized in a mixed gel of gellan and xanthan on beads and are then introduced in a single or multi-stage chemostat (continuous culture system) simulating the biofilm typically forming in the GI tract. This system allows the microorganisms to adhere. At the same time, however, it lacks the key point that specifically characterizes the mucosal biofilm namely, the anaerobic conditions prevailing at the top of the biofilm and microaerophilic conditions prevailing at the base of the biofilm.
Probert and Gibson (2004) provided a similar device with a framework of mucin beads encased within a dialysis membrane. The system is inoculated with fecal samples and water and metabolites are removed by osmosis using a solution of polyethylene glycol.
Finally, Macfarlane et al. (2005) developed a two-stage continuous culture system, simulating the proximal and distal colon, and used sterile porcine mucin gels in small glass tubes to determine how intestinal bacteria colonize and degrade mucus. These tubes can be placed in a fermentor simulating a specific area of the GI tract and removed over a period of 48 h for further analyses of on the biofilm.
These systems allow the microorganisms to adhere but do not offer the opportunity of studying the gut biofilm formation (Lebeer et al., 2007) and the host-microbial interaction under continuous simulated conditions.
Simulation of the Host-Bacteria Interaction
None of the aforementioned models simulating the GI tract has an adequate device to study the mechanisms of bacterial adhesion in response to the host signals and the reciprocal cross-talk. Previous studies use either germ-free animals (mainly rodents and, more recently, zebrafishes—Cheesman and Guillemin, 2007) or cell culture experiments (mainly Caco-2 or HT29 cells).
Animal studies demonstrated that vertebrates possess a broad scala of preserved interactions with the microbes with which they co-evolve (Cheesman and Guillemin, 2007) in particular when maintaining gut epithelial homeostasis, however, mechanistic studies are not always possible.
The in vitro use of cell lines can be limited by the fact that these cells do not produce a mucus layer (Caco-2 cells) or by the fact that pure microbial cultures or only a mix of few strains can be tested, for reasons of cytotoxicity. Cell cultures are very sensitive to co-incubation with mixed microbial slurries, thus limiting the incubation time and the adaptation of the host and the microbial metabolism.
One interesting model has been developed by Laube et al. (2000) to simulate the sequential metabolism of chemicals by the liver and the intestinal microbiota. Here, in a double chamber system, hepatocytes are cultivated as a monolayer on a membrane while, in the anaerobic compartment, the fecal microbiota are present in suspension. The exchange of metabolites can take place across the permeable membrane.
Another study has been conducted by Parlesak et al. (2004), investigating the interaction between human mononuclear leucocytes and enterocytes during challenge with a single bacterial species using compartmentalized transwell cell culture systems.
The transwell cell culture system has also been used by Linden et al. (2007), in which human gastrointestinal epithelial cell lines (e.g. MKN1, MKN7, Caco-2, . . . ) were grown on the apical side of a transwell and subsequently cocultured with different microbial strains.
Although the above described systems are very useful for short-term experiments, they are generally not suited to study the complex properties of the intestinal microflora over long-term studies, due to the cytotoxicity of the microbial cells towards the human cell layer.
In Vitro Model of the Present Invention
An in vitro test that does not represent the actual in vivo complexity will be unreliable when the results are extrapolated to an in vivo GIT situation (Pedersen and Tannock, 1989). Mucosal bacterial communities in the GI tract are difficult to study in vivo and biopsies are usually obtained from diseased individuals. Consequently, the data available may not provide a true indication of a normal mucosal diversity (Macfarlane and Dillon, 2007). Furthermore, an adequate device to study the mechanisms of bacterial adhesion in response to the host signals and their reciprocal cross-talks is not yet available.
Thus, there is a need for in vitro models that                reflect the in vivo GIT situation,        mimic the relevant environmental conditions of a mucosal layer,        can be adapted to a continuous system        allow the study of the adherence, colonization, composition and metabolic activity of the mucosal microbial population over a longer time-frame,        allow the formation of a mucosal biofilm with specific (in particular anaerobic or aerobic) conditions prevailing at the top of the biofilm and microaerophilic conditions prevailing at the base of the biofilm,        provide the possibility to perform experiments both with mono-cultures and with mixed and hence more relevant microbial communities, and/or        allow to evaluate the host-microbiota interaction and the consequent reciprocal adaptation.        
The present invention describes different models to study microbial adhesion to mucosal surfaces. In particular the models of the present invention comprise 2 compartments separated by a semi-permeable membrane. Said membrane, on the luminal side being coated with an artificial mucus layer to which microorganisms applied in the luminal compartment are allowed to adhere. The use of said artificial mucus layers is advantageous compared to the use of mucus layers formed by epithelial cells, since direct interaction and as such also cytotoxicity between the microorganisms and the epithelial cells is prevented, allowing long-term analyses. Nevertheless, epithelial cells and/or other cell types can be grown in the basal compartment of the module and secretion products of these cells as well as of the microorganisms are allowed to diffuse through the membrane and artificial mucus layer in both directions. Furthermore, the use of two separate compartments allows the establishment of different oxygen pressures on both sides of the membrane. By regulation of the oxygen pressure in the individual compartments, optimal conditions for the aerobic epithelial cells in the basal compartment as well as for the anaerobic microorganism in the luminal compartment can be established. As will become evident from the examples hereinafter, the oxygen gradient across the semi-permeable membrane results in microaerophilic conditions at the luminal side of the artificial mucus layer that closely mimics the corresponding in vivo situation for the adhesion of microorganisms to said layer. Finally, the models allow an establishment of shear stress in the compartment containing the microorganisms, which is very important to mimic the in vivo situation. The combination of these features clearly distinguishes the model of the present invention from the models described in the prior art and provides a novel gastrointestinal tract model that closely mimics the corresponding in vivo situation.